My personal assistant in Claude
I give it context and criteria to prepare my tasks.
I'll show you, step by step, how to have AI prepare the work you repeat every week, whatever you do. AI prepares. You review and decide.
Free · One real task in 30 minutes · No purchase required
Draft replies to my customers in my tone, following my rules
Works with the tools you already use.
From an idea to something I can use
I’ll show you how to build it with your tools and review each result before using it.
I give it context and criteria to prepare my tasks.
I explore my data in Power BI and check the figures.
I prepare a draft with Copilot and apply my judgement.
I organise ideas with ChatGPT or Gemini and add my voice.
I describe what I need and review the site step by step.
I connect the steps of a task and decide what to approve.
Here’s a hypothetical week: times include review and are not a promise of savings.
Gather data and prepare my report3 h
Write my posts and scripts2 h
Sort emails and draft replies2 h
Gather data and prepare my report1 h
Write my posts and scripts1 h
Sort emails and draft replies1 h
Hypothetical example: 7 h down to 3 h, review included
What do you do?
Illustrative times per task, not guaranteed results. They include your review and depend on the context.
Before · Compare transactions one by one
After · The agent matches data; you review discrepancies
Before · Gather and check spreadsheets
After · Checklist and draft close for validation
Before · Copy figures into a report
After · A sourced draft for you to verify
Before · Check due dates and write emails
After · Prioritised list and reminders for approval
Before · Reread notes and transcripts
After · Summary and next steps for you to confirm
Before · Write each reply from scratch
After · Contextual draft for review
Before · Tag and assign tickets manually
After · Suggested classification under your supervision
Before · Find new answers across tickets
After · Suggested updates for you to validate and publish
Before · Open documents and copy data
After · Extraction with references for checking
Before · Type records into another tool
After · Records prepared for review and import
Before · Write procedures from notes
After · Structured draft for you to complete
Decisions about people are always human. AI helps organise information; it does not decide who gets hired.
Before · Read CVs and organise relevant experience
After · Summaries against explicit criteria for human evaluation
Before · Prepare each checklist from scratch
After · Onboarding plan for you to personalise
Before · Look up policies and write answers
After · Sourced answer for you to validate
Before · Match availability manually
After · Proposed schedule for you to agree
Before · Match orders against stock
After · Discrepancy list for you to verify
Before · Collect emails for each issue
After · Summary and follow-up for your approval
Before · Copy logs into a daily summary
After · Report with alerts for you to review
Before · Gather notes and draft an offer
After · Contextual proposal for you to adjust
Before · Find the last conversation and write
After · Suggested follow-up for approval
Before · Update records after meetings
After · Changes extracted from notes for validation
Before · Consolidate metrics from multiple sources
After · Updated dashboard whose data you cross-check
Before · Export and match sheets manually
After · Agent prepares matching for your review
Before · Prepare the financial report manually
After · Draft and Power BI metrics for validation
Automated report
September financial report

About me
Computer engineer · AI and data analytics
I've spent 6 years in data and BI, and today I build with AI every day: Claude and Claude Code, OpenAI's Codex, other LLMs and AI agents that work as a team. What I teach here is what I use on real projects, adapted so you can apply it in your role without coding.
No hype: I'll show you what works, where it fails and when AI isn't worth it.
Start with a real task
Free, with immediate access to the kit. No purchase required.
The syllabus
Learn to ask for what you need, with the data and criteria that make an answer useful.
Design assistants for your recurring weekly tasks, with clear instructions and boundaries.
Break a task into steps, define what the agent needs and when it should ask for your help.
Connect email, calendars, Excel, ERP, CRM and social media to your everyday tasks.
Check responses, protect data and learn when not to use AI. You decide what gets approved.
Identify tasks in your role, choose where to start and evaluate what works before expanding.
Illustrative preview of the course
Enrollment 20–26 November · €97
Secure payment with Stripe · VAT included
Available methods are shown at checkout depending on your device and country.
20–26 November 2026 · Madrid timeNo. You’ll learn to give instructions, use assistants and review their responses without coding.
If your job has tasks that repeat, yes: finance, customer support, back office, HR, operations, sales, social media or management. You pick one task from your job and turn it into an assistant that prepares the work for you to review.
The ones you already use or can use at work: Claude, ChatGPT, Microsoft Copilot or Gemini, together with Excel, email, your ERP or Power BI. Some features require paid plans from those tools.
Opens on 20 November 2026 at 00:00 and closes on 26 November at 23:59, Madrid time. Launch price: €97. Then enrollment closes until the next intake, with no date announced.
Yes. For teams I design training by department, automation with AI agents and Power BI analytics: tell me about your case on the business page. See training for businesses →
Marcos Valle is a computer engineer, data analyst and AI engineer with 6 years of experience. He builds AI-powered ERPs, data systems and AI agents, with experience on projects for large multinationals, startups and SMEs, and teaches professionals in any role to automate their work with AI, no coding required.
An assistant answers and drafts content when you ask. An AI agent chains several steps on its own (reading an email, checking a spreadsheet, drafting the reply), uses tools within the limits you define and asks for your review before closing the task.
A skill is a reusable package of instructions, examples and criteria that the assistant loads when a task needs it, so the AI does a task you repeat every week equally well every time.
MCP (Model Context Protocol) is an open standard for connecting AI assistants such as Claude to tools and data: email, calendar, databases or an ERP. It lets the AI work with your real information instead of hand-pasted text.
It depends on the tool, the plan and how it's configured. In the course you'll learn which data not to share, how to review what the AI does and when not to use it; the final decision is always yours.
An assistant template for your role, a checklist for reviewing the AI's output and the steps to automate your first task in about 30 minutes. You can download it immediately and doesn't commit you to the course.
With the free kit you set up your first task in 30 minutes. If you want to go further, the course is open 20–26 November for €97.